imgVOID / autograding-api

A simple Docker sandbox example and a ready-to-use autograder API. Based on asynchronous FastAPI and disposable Docker containers. Three modes of operation: Python subprocess, Python Docker SDK and Python on Whales.

Autograding API

🚀 Fast and asynchronous API for automated grading of code assignments.

1. FastAPI provides support for OpenAPI 3.0 and Swagger.
2. Separate disposable Docker containers to inspect and grade potentially unsafe user input.
3. Three working modes: "Python Subprocess", "Python on Whales" and "Python Docker SDK"
4. Fully RESTful (CRUD) API for use with a custom frontend.
5. Authentication based on the Bearer JWT.
6. Asynchronous ORM SQLAlchemy support.

API sections:

• tasks: CRUD for the programming assignments management.
• topics: CRUD for the tasks topics management.
• checks: validating user input by running securely in a disposable Docker container.
• users: user management system.

Requirements:

1. FastAPI.
2. Python Docker SDK (optional).
3. Python on Whales (optional).
4. Pydantic.
5. Aiofiles.
6. Pytest and pytest-asyncio.
7. Uvicorn.
8. Databases (Python module).
9. Httpx.
10. Python-jose.
11. Passlib.
12. Slowapi (optional).

TODOs:

1. Finish the "topics" section by analogy with the "tasks" section.
2. Improve the sandboxing algorithm using Docker volumes or Docker bind mounts.